On Thu, Apr 14, 2022 at 10:44:48PM +0000, Alexander Lobakin wrote:
> When CONFIG_PERF_EVENTS is not set, struct perf_event remains empty.
> However, the structure is being used by bpftool indirectly via BTF.
> This leads to:
>
> skeleton/pid_iter.bpf.c:49:30: error: no member named 'bpf_cookie' in 'struct perf_event'
> return BPF_CORE_READ(event, bpf_cookie);
> ~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~
>
> ...
>
> skeleton/pid_iter.bpf.c:49:9: error: returning 'void' from a function with incompatible result type '__u64' (aka 'unsigned long long')
> return BPF_CORE_READ(event, bpf_cookie);
> ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>
> Tools and samples can't use any CONFIG_ definitions, so the fields
> used there should always be present.
> Move CONFIG_BPF_SYSCALL block out of the CONFIG_PERF_EVENTS block
> to make it available unconditionally.
Urgh, this is nasty.. did you verify nothing relies on that structure
actually being empty?
Also, why are we changing kernel headers to fix some daft userspace
issue?
> Fixes: cbdaf71f7e65 ("bpftool: Add bpf_cookie to link output")
> Signed-off-by: Alexander Lobakin <alobakin@xxxxx>
> ---
> include/linux/perf_event.h | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
> index af97dd427501..b1d5715b8b34 100644
> --- a/include/linux/perf_event.h
> +++ b/include/linux/perf_event.h
> @@ -762,12 +762,14 @@ struct perf_event {
> u64 (*clock)(void);
> perf_overflow_handler_t overflow_handler;
> void *overflow_handler_context;
> +#endif /* CONFIG_PERF_EVENTS */
> #ifdef CONFIG_BPF_SYSCALL
> perf_overflow_handler_t orig_overflow_handler;
> struct bpf_prog *prog;
> u64 bpf_cookie;
> #endif
>
> +#ifdef CONFIG_PERF_EVENTS
> #ifdef CONFIG_EVENT_TRACING
> struct trace_event_call *tp_event;
> struct event_filter *filter;
> --
> 2.35.2
>
>
